Merge pull request #212 from Herringway/fix-uri-escapes

fix UTF-8 decoding in URI sequences
merged-on-behalf-of: Cameron Ross <elpenguino@gmail.com>

source/dyaml/scanner.d

@@ -1973,9 +1973,16 @@ final class Scanner
                     ++nextChar;
                 }
                 else
+                {
+                    try
                     {
                         c = decode(bytes[], nextChar);
                     }
+                    catch (UTFException)
+                    {
+                        return size_t.max;
+                    }
+                }
                 reader_.sliceBuilder.write(c);
                 if(bytes.length - nextChar > 0)
                 {
@@ -2016,8 +2023,8 @@ final class Scanner
 
                     uint digit;
                     if(c - '0' < 10)     { digit = c - '0'; }
-                    else if(c - 'A' < 6) { digit = c - 'A'; }
-                    else if(c - 'a' < 6) { digit = c - 'a'; }
+                    else if(c - 'A' < 6) { digit = 10 + c - 'A'; }
+                    else if(c - 'a' < 6) { digit = 10 + c - 'a'; }
                     else                 { assert(false); }
                     b += mult * digit;
                     mult /= 16;
@@ -2028,6 +2035,13 @@ final class Scanner
             }
 
             bytesUsed = getDchar(bytes[0 .. bytesUsed], reader_);
+            if(bytesUsed == size_t.max)
+            {
+                error(contextMsg, startMark,
+                        "Invalid UTF-8 data encoded in URI escape sequence",
+                        reader_.mark);
+                return;
+            }
         }
 
 

source/dyaml/test/resolver.d

@@ -16,6 +16,12 @@ import std.string;
 import dyaml.test.common;
 
 
+string construct(ref Node node) @safe
+{
+    return node.as!string;
+}
+
+
 /**
  * Implicit tag resolution unittest.
  *
@@ -37,7 +43,13 @@ void testImplicitResolver(string dataFilename, string detectFilename) @safe
     }
 
     correctTag = readText(detectFilename).strip();
-    node = Loader.fromFile(dataFilename).load();
+
+    auto constructor = new Constructor;
+    constructor.addConstructorScalar("tag:example.com,2000:app/tag🤔", &construct);
+    auto loader = Loader.fromFile(dataFilename);
+    loader.constructor = constructor;
+
+    node = loader.load();
     assert(node.isSequence);
     foreach(ref Node scalar; node)
     {

test/data/uri.data

@@ -0,0 +1,3 @@
+%TAG !e! tag:example.com,2000:app/
+---
+- !e!tag%F0%9F%A4%94 baz

test/data/uri.detect

@@ -0,0 +1 @@
+tag:example.com,2000:app/tag🤔